01 · What breaks
Design for the failure modes, not only the happy path.
Graduation programmes run for long periods and serve households close to the edge, so small design choices compound. These four problems are the ones that turn a careful approach into a single score nobody trusts.
One composite score hiding trade-offs
A household scoring 7 out of 10 might have steady income and no food security, or the reverse. Averaging dimensions hides the one that matters most to the family. Show each dimension separately against its threshold, so reviewers see where a household is strong and where it is still exposed.
Income estimates treated as precise
Income in households with seasonal work, livestock and remittances moves month to month and is usually recalled, not recorded. Store it as a band with its source and season, next to harder evidence such as assets owned or a savings balance. A precise-looking figure invites decisions it cannot support.
Household composition changes ignored
When an earning member migrates, a widowed daughter returns or a baby arrives, the household's position shifts. If the system keeps only the baseline composition, progress figures end up comparing different households. Record membership changes as dated events and prompt a reassessment of the dimensions they affect.
Graduation used to end support automatically
If reaching a threshold closes the case and stops coaching visits, a household hit by a shock the following month has nobody watching. Keep graduation as a decision recorded by a review, followed by post-graduation check-ins, and let a household return to active support without starting over.
02 · Minimum viable model
Record each household's path across dimensions, services and shocks.
The graduation approach began with BRAC's Targeting the Ultra Poor programme in Bangladesh in 2002, and programmes since have combined several supports over a fixed period. Whatever your package, these six records show what each household received, what changed and what set it back.
| Dimension | Evidence, stronger to weaker | Illustrative threshold (set your own) |
|---|---|---|
| Food security | Ration records or observed meals, then recall of the past week | Two meals a day through the past month |
| Income sources | Sources confirmed on a visit, then self-report | At least two active sources, one of them non-farm |
| Savings | Passbook or SHG ledger balance, then self-report | Regular deposits for six months |
| Productive assets | Asset seen and in use, then purchase record | Transferred asset still owned and productive |
| Health and scheme access | Card or entitlement seen, then self-report | Enrolled in the schemes the household is eligible for |
| Confidence and planning | Coach observation with notes | Household can describe its plan for the next lean season |
The thresholds above show the structure only. Agree each one with your MEL lead and field teams before the baseline. Where a programme also uses an income line, such as the Lakhpati Didi definition of a woman SHG member earning at least ₹1 lakh a year, treat it as one dimension among several, not the whole decision.
03 · Implementation sequence
Move through reviewable gates.
Agree the dimensions before building anything, because every form, view and review depends on them. If the programme's logic is not yet expressed as records, the Theory of Change to data model guide covers that step.
Agree dimensions and evidence
Programme, MEL and field leads agree the dimensions, the evidence accepted for each and the threshold that counts as met. Write them as an indicator spec with a definition and a source. This document governs every later review, so changes go through the same people and carry a date.
Create a stable household identity
Assign each household an ID at enrolment and link members individually with their relationship to the head. Check for duplicates across existing projects, because the same family may already be in an SHG or skilling programme. One household linked to several projects keeps cost per graduated household honest.
Record the service pathway
Each component of the package (asset transfer, consumption support, coaching visits, savings linkage, health enrolment) is a dated event against the household, with the project code that pays for it. The team can then see who missed a component and calculate cost per household without a year-end spreadsheet.
Review progress and shocks together
Coaching visits record movement on each dimension and any shock: illness, crop loss, a death, a new debt. The monthly review shows both in one view, so a household that slipped because of a hospital bill gets support rather than a failing mark. Shocks use a short coded list plus a note.
Conduct a documented graduation review
At the end of the cycle, a review panel looks at each household's dimensions, evidence dates and shocks, then records graduate, extend or change plan, with reasons. Exceptions to the thresholds are allowed but written down. The system prepares the review packet; the panel signs the decision.
Follow up for sustainability
Schedule check-ins at agreed intervals after graduation, using the same dimension questions as the baseline so results compare. Households that slip are offered renewed support. The programme lead checks follow-up coverage before reporting sustained outcomes, and unreached households stay in the denominator.
04 · Evidence
Measure whether the work became clearer.
A graduation count means little without knowing how current the evidence is and how many households were seen again later. These measures keep the headline honest. The wider model sits in our livelihoods playbook.
05 · Responsible automation
Let systems prepare. Keep consequence with people.
Systems can support
- Dimension-by-dimension progress views
- Service receipt and cost per household
- Shock logging with reassessment prompts
- Review packets and post-graduation schedules
People should decide
- Graduation and continuation decisions
- Interpreting household shocks
- Resolving conflicting income evidence
- Changing the support plan
06 · Common questions
Practical questions teams ask.
Questions livelihoods teams ask when moving a graduation programme off paper.
What is a household graduation tracking system?
It records each household's baseline across agreed dimensions, the services it receives, the shocks it faces and repeat assessments over the programme period. It then supports a documented review that decides whether the household has graduated. The aim is to see progress on each dimension clearly, not to produce one score that makes the decision for the team.
What dimensions should graduation criteria cover?
Most programmes look at food security, income sources, savings, productive assets, access to health and government schemes, and some measure of confidence or planning. The exact set and thresholds should come from your programme design and local context. Each dimension needs a definition, accepted evidence and a threshold, all agreed before the baseline is taken.
How do you handle households hit by a shock during the programme?
Record the shock as a dated event with a type and a note, and review it alongside progress at the next meeting. A household set back by illness or crop loss may need extra support or a longer timeline. Shocks should change the plan, not count against the household or the coach in performance reports.
Should graduation happen automatically when thresholds are met?
No. Thresholds show a review panel where to look, but the decision needs people who know the household, its risks and what the evidence leaves out. Record the decision, the reasons and any exceptions. Automatic closure risks ending support just before a lean season or after a change in the family that nobody recorded.
How long should households be followed after graduation?
Long enough to see whether gains last through at least one difficult season, and at intervals the programme can actually cover. Use the same dimension questions as the baseline so results compare. Report sustained graduation with follow-up coverage beside it, and keep households you could not reach in the total instead of quietly dropping them.